[comp.lang.ada] writing operating systems in ADA

wytten@umn-cs.CS.UMN.EDU (Dale R. Wyttenbach) (10/19/89)

I'm taking an ADA class this quarter and in order to get an 'A' I have
to write a paper.

I would like the topic to be either "writing operating systems in ADA", or 
"why ADA is not appropriate for writing operating systems".  I'll know
which when I'm through writing it ;-)

I need help finding sources for such a paper.  If you know of any articles
on this topic, please send me the reference (bibliography-style)

Much appreciated

dale


 Dale Wyttenbach		     |		...rutgers!umn-cs!wytten
 wytten@umn-cs.cs.umn.edu	     |		   wytten@umnacvx.bitnet
Computer Science Department Systems Staff--University of Minnesota, Minneapolis

phil@ingr.com (Phil Johnson) (10/20/89)

Intel attemped to build an Ada-based machine (using Ada to write the operating
system) but found that it was pretty much a wash. The restricted lexical 
elements could not support general languages which used characters not allowed
in Ada (APL comes to mind - ain't no Thumbnail in Ada).  I do not have any 
references to the Intel machine, but perhaps someone on the net does.



-- 
Philip E. Johnson                    UUCP:  usenet!ingr!b3!sys_7a!phil
MY words,                           VOICE:  (205) 772-2497
MY opinion!

billwolf%hazel.cs.clemson.edu@hubcap.clemson.edu (William Thomas Wolfe, 2847 ) (10/21/89)

From article <6991@ingr.com>, by phil@ingr.com (Phil Johnson):
> Intel attemped to build an Ada-based machine (using Ada to write the operating
> system) but found that it was pretty much a wash. The restricted lexical 
> elements could not support general languages which used characters not allowed
> in Ada (APL comes to mind - ain't no Thumbnail in Ada).  I do not have any 
> references to the Intel machine, but perhaps someone on the net does.

    However, a joint venture between Intel and Siemens (Biin) has
    has considerably better results: Biin markets multiprocessor
    systems designed for high-performance online transaction processing,
    and their Biin/OS operating system is written entirely in Ada, as
    are most system services.  Reference: 1-800-252-Biin in North America,
    (49) 911 5219 0 in Europe. 


    Bill Wolfe, wtwolfe@hubcap.clemson.edu

karl@grebyn.com (Karl Nyberg) (10/22/89)

In article <6842@hubcap.clemson.edu> billwolf%hazel.cs.clemson.edu@hubcap.clemson.edu writes:
>    However, a joint venture between Intel and Siemens (Biin) has
>    has considerably better results: ...

If BiiN (note spelling) were having such considerably better results, why
would Intel and Siemens be selling it off?

-- Karl --

farber@linc.cis.upenn.edu (David Farber) (10/22/89)

There are many reasons one can sell off a enterprise , sometime its
not only due to technical problems or sales. Note that GM sold
off EDS and EDS was not a failure.

David Farber; Prof. of CIS and EE, U of Penn, Philadelphia, PA 19104-6389 Tele:
215-898-9508(off); 215-274-8292 (home); FAX: 215-274-8192;  Cellular:  302-740-
1198 "The fundamental principle of science, the definition almost, is this: the
sole test of the validity of any idea is experiment." -- R. P. Feynman

Jerome_V_Vollborn@cup.portal.com (10/23/89)

I think the poster is refering to the Intel 432.  I don't remember the 
operating system being written in Ada but the 432 did support object 
in the hardware.  Each processor was considered an object and had 
a stored object description.  See Elliot I. Organick' book "A
Programmer's View of the Intel 432 System" (McGraw-Hill Book Company, 
1983, ISBN 0-07-047719-1).

The major problems with the 432 were marketing and technical.  The 
marketing problem was that the 432 did not look like an 8080 (1/2 :-).
The technical problem was that the 432 was a bus hog.  When it went 
to request data from memory, it sent a processor number, an address 
and a length.  The return packet was the processor number, address, 
length and data.  Too expensive for Multibus I to handle.

				Jerome Vollborn
				(Jerome_V_Vollborn@cup.portal.com or
				 uunet!lci386!jerome)

chittamu@umvlsi.ecs.umass.edu (Satish Kumar .C) (10/23/89)

In article <6991@ingr.com> phil@ingr.UUCP (Phil Johnson) writes:
>Intel attemped to build an Ada-based machine (using Ada to write the operating
>system) but found that it was pretty much a wash. The restricted lexical 
>elements could not support general languages which used characters not allowed
>in Ada (APL comes to mind - ain't no Thumbnail in Ada).  I do not have any 
>references to the Intel machine, but perhaps someone on the net does.

I believe the Microprocessor that Intel developed as an Ada based
machine was the iAPX 432. You might want to check on that though. I
have been told (but I make no guarantee) that there was an article in
IEEE computer sometime around mid 1985. 

tom@telesoft.com (Tom Quiggle @sation) (10/24/89)

In article <16422@umn-cs.CS.UMN.EDU>, wytten@umn-cs.CS.UMN.EDU (Dale R. Wyttenbach) writes:
> 
> I would like the topic to be either "writing operating systems in ADA", or 
> "why ADA is not appropriate for writing operating systems".  
> ... If you know of any articles on this topic, please send me the 
> reference (bibliography-style)
> 

You will want to read the following:

%A D. Keeffe
%A G. M. Tomlinson
%A I. C. Wand
%A A. J. Wellings
%T Pulse: An Ada-based Distributed Operating System
%I Academic Press
%O Volume 26 in A.P.I.C. Studies in Data Processing
%C London
%D 1985

The book is a comprehensive discussion of the design and implementation
of a distributed operating system in (and for) Ada.  It includes an 
excelent, though no longer up-to-date, bibliography.

Tom Quiggle

-- 
--
Tom Quiggle                     TeleSoft                  (619)457-2700x158

UUNET: ucsd!telesoft!tom                   ARPA:  quigglet@ajpo.sei.cmu.edu

    bumper sticker on stealth bomber:
      "if you can read this, we wasted 50 billion dollars".
			  -- David Letterman 7/27/89

callen@inmet (10/24/89)

The entire operating system for the Rational R1000 is written in Ada; the
underlying hardware is built to run Ada, with intrinsic support for such
high level concepts as rendezvous. You might give them a call; they are in
Santa Clara, CA.

-- Jerry Callen
   ...!uunet!inmet!callen

kannan@babcock.cerc.wvu.wvnet.edu (R. Kannan) (10/25/89)

PULSE: An Ada based DISTRIBUTED OS, was designed and implemented in UK.
It has been around since 1985. It is more like UNIX. Has all the interface for
UNIX.
There is a book on PULSE.
--kannan

snidely@nosun.UUCP (David Schneider) (11/02/89)

In article <6991@ingr.com> phil@ingr.UUCP (Phil Johnson) writes:
>Intel attemped to build an Ada-based machine (using Ada to write the operating
>system) but found that it was pretty much a wash. The restricted lexical 
>elements could not support general languages which used characters not allowed
>in Ada (APL comes to mind - ain't no Thumbnail in Ada).
>

I'm sure that most OS writers would be suprised to know that
they had to program in a language which supports "thumbnail"
in *it's* lexical elements for APL to run on that OS.

The 432 died because it was too slow.  The OS was less a factor
than bus problems (but performance improved scalably with additional
processors).

The second generation project also used Ada for the OS,
with a compiler for the 80960 that ran fast and produced good code.
This project, a joint venture with Siemens and yclept "BiiN",
failed for market/project management reasons, not because of
any technical problems with using Ada.

In fact, I would go with Ada as an OS language again.
The important parts are (to the language user) about the
same size as Pascal, and the package concept is useful
to design whether you're doing SASD, JSD, or OOD.
(I will side with Grady Booch and claim that Ada is sufficient
for an object-based design, even if the slim support of
inheritance is a liability).


				Dave Schneider
				late of BiiN